- This repository includes a tensorflow implementation of SiamRPN[1]
- The best AUC score on OTB100 of this repository so far: 0.602 (the AUC score in the paper[1] is 0.637)
- less training videos(unable to download YouTube-BB 😵...)
- DET2014+VID2015+LASOT+GOT10k(about 20k videos)
- SiamRPN[1] use YouTube-BB+VID2015(about 100k videos)
- more data augmentation
- random mixup [2]
- random image blur & color jittering
Add more details ... (TODO)
- Training dataset is more important than anything else
- Use pretrained feature extraction and train the network from deep layer to shallow layer step by step
- first train deeper layers with larger learning rate
- and then train the whole network with smaller learning rate
- Models trained from scratch always perform worse (best AUC score on OTB100: 0.564) in my experiment even with longer training time(5 times), this might be attributed to limited training dataset
- Achieve the AUC score in the paper[1]
- To speed up training by replace tf.py_func & numpy operation with pure tf implementation
- Multi-GPU Training & load dataset with lmdb lmdb dataset download(about 120G) code: dn58
- Convert pytorch pretrained model to initialize embedding function
- learning rate warmup
- Train network with static images and dynamic videos
- Add time decay option to weight loss[3] (a slight improvement)
- Add random Mixup[2]
[1] High Performance Visual Tracking With Siamese Region Proposal Network (CVPR 2018)
[2] Bag of Freebies for Training Object Detection Neural Networks
[3] Learning Attentions: Residual Attentional Siamese Network for High Performance Online Visual Tracking